24/192 is IMO what they should have targeted instead of 24/96...while still "ok" the extra uplift/capability difference between 96 and 192 (with proper source) is a huge difference.
also they should have supplied more info on how much "power" this DAC has I imagine because it is powered by USB alone it will not have a bunch of extra power to drive higher impedance headphones, also, just to point it out, while it seems "neat" and might offer a few things others do not, that is a hell of an asking price IMO when one is able to get stand alone DC/Amplifier for a fraction of the price.
because it is steel series likely they require their drivers (which are at best "passable" if they are not wanting to update themselves all the damn time lol...yes I had experience with steel series stuff)
Apart from chasing specs - what is the benefit you see in 96khz? The ability to play some ultrasonic frequencies that could potentially degrade your listening experience (if your headphones don't like it etc etc)?
48khz is way more than enough frequencies (thanks to Nyquist - 24khz) for you to listen to, and (at least for me) about ~6khz of that I can't really hear anyway.
USB3 can supply ~ 0.9amp - which should be more than enough to make those higher frequencies irrelevant to you quickly enough :P
24/96 doesn't need a driver on Windows, as the stock UAC2 driver covers it.
I too prefer hardware that supports at least 24/192. While the extreme higher frequencies may largely be inaudible, the lower noise floor and higher dynamic range will benefit anyone with decent headphones. And the more capable hardware required to support 24/192 tends to make require higher quality components, which generally benefits ALL sound, except poorly encoded or highly compressed sources.
Frequencies above ~20kHz are not "largely inaudible" - they are completely inaudible. Furthermore reproducing ultrasound frequencies may introduce unwanted artifacts in the audible range due to non-linearities of the speakers etc..
Using sample frequencies of 96/192 ksps is marketing bullshit. And frankly, since when has "gaming" audio been synonymous with high-end Hifi anyway?
Have you ever listened to a 24/192 FLAC of uncompressed music (and by this I mean music that a producer has not artificially compressed) with good headphones on quality hardware? It's not all about high frequency, it's also about being able to hear low level sound that would be obscured by chips unable to produce the dynamic range need to hear EVERYTHING.
Gentlemen - 96kHz refers to the sampling rate, not the sound frequency. Obviously there is zero value in capturing 96kHz frequencies that your dog can't hear. That's not what 96kHz is in reference to.
A better way to understand it is to think about 3D printing. Have you seen 3D prints done on cheap printers? You know how you can see the steps in the side of the object as it builds level by level on the 3D print, making the printed object looks crappy and a lot rougher in texture than the original design. It can't print a continuous object, instead it has to break it down into slices and then express it that way. Well, one difference between cheap 3D printers and nice ones is the nice ones can use really fine slices, so fine that if you stand at a reasonable distance you might not notice they're there and it approximates the original design.
This is what 48/96/192kHz is talking about. Your favorite song that includes audible sounds from 20Hz to 20kHz, how finely do you want the compressed, digitized "print" or copy of that song to slice it? The finer it's sliced, the better it sounds, to a point obviously. See "Sampling" on Wikipedia.
tl;dr and tl;dw: sampling rate defines the maximum frequency you can record (specifically sampling rate/2 == max, so 44.1kHz sample rate of a CD gives you 22.05kHz max, for example). It has absolutely no bearing on "resolution" of sound, because we can already perfectly represent any sound any human can hear.
Very interesting links, and they do cast doubt on my 3D printing analogy, but the links also teach sampling rate does not equal the sound frequency, which is what my overall point was. A 96kHz sampling rate does not mean the sound frequency will be extended out to 96kHz. Your first link had a link to an interesting article: http://lavryengineering.com/pdfs/lavry-sampling-th... . The Nyquist theory, from what I can read, describes the number of slices you'll need to perfectly describe a sound or frequency. Adding more slices adds nothing to the presentation. In this way, the analogy to screen resolution isn't terrible, in that we can reach a point with resolution that adding more pixels to a phone screen or a PC display adds zero value or additional data but comes at a great cost, in file size, in processing requirements, in hardware requirements, in power requirements. I'm not proposing a certain sampling rate, but I do know the sampling rate does not equal the sound frequency contained by a file. If that were the case, we could be happy with 24kHz sampling, but give that a try sometime and see how it works out for you compared to 48kHz.
How many audio devices have a frequency output higher than 22Khz? Very few.
I've seen hi-res 'buffs' going on about how hi-res is amazing then i ask what speakers they run it all through...oops. Bit rate probably helps more with resolution than frequency. They should have just upped Red Book to 24bit and left it at that. 18 or 20bit would have probably been fine.
Human hearing is generally accepted to be 20Hz to 20kHz, though some argue that true golden ears can hit 21, or even 22kHz (I haven't heard of any yet, though...). It doesn't matter that there are plenty of speakers, DACs and amps that can render higher (albeit very non-linearly, and we know that cause we can measure it), if you can't hear it.
Remember, you have to account for the noise floor - 20dB(SPL) in a studio, 30-50 dB(SPL) for more typical surroundings; and the threshold of pain - ~120dB(SPL). That means that the maximum theoretical dynamic range you can reach is 100dB. 16bits lets you encode 96dB of dynamic range without using any fancy tricks.
In practice, the human ear has much lower dynamic range because human hearing dynamically adjusts to compensate for loudness. Combine with not wanting music to be "too loud", one can reasonably conclude that the 96dB dynamic range of 16bit audio is more than enough to capture any human-perceptible audio. Until the end of time.
That said, there are reasons for having higher bit depths, and that would be mixing, production and mastering, where higher bit-depths just make things more convenient. You can, of course still work entirely in INT16, but you need to take some care with effects, clipping and distortion. But working in INT24, or FP32 (in particular) is just so much more convenient, so that's what you see in production.
>They should have just upped Red Book to 24bit and left it at that. 18 or 20bit would have probably been fine.
No that would have been stupid because the human ear is incapable of discerning higher than 16bit fidelity. It is literally mechanically impossible for the ear to discern greater fidelity, there is no higher quality.
Of course many people including audiophiles are going to explain that this is wrong because they don't understand sampling and because they don't understand that confirmation bias is automatic and unconscious.
Bit depth has a much larger impact than frequency. The audible difference between 96 and 192 Kilohertz is very minor. You'd be much better off increasing the bit depth to 32 bit floating point.
The ONLY place you need anything higher than that is when you're actually mastering or mixing audio. That would be handled internally by WASAPI/ALSA/OSS/PulseAudio/JACK, not by the DAC. For pure audio storage and output, 16/44.1 is all you need, although DVD and BR have pushed the standard a little up to 24bit/48kHz (somewhat unnecessary, but not actually harmful).
What a bunch of F***ers. I bought the bundle when it first came out and you can't use any other headphones with the DAC. Now they release one 3 months later that you can use any headphones. People don't buy their headphones, don't listen to the reviews.
I imagine the standalone version is just going to come with a cable to plug in the 3.5mm headsets and SteelSeries sells their cables on their site. Basically I would expect to see the opposite of this appear on the site soon:
It's nice the DAC is standalone now. I'm pretty excited about that because I can pair it up with any old set of earbuds or headphones without blowing $70+ on overpriced Arctis ones when I'm fine with a $10 throw away set off Amazon or out of the local Wally World.
The DAC reminds me of the Soundblaster G5 I have, which I think is dying on me as sometimes it randomly disconnects and connects. I wonder if you can listen to both optical and PC audio at the same time on this one. That is the killer feature that has kept me from trying other dac/headphone amp solutions.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
23 Comments
Back to Article
Dragonstongue - Tuesday, August 28, 2018 - link
24/192 is IMO what they should have targeted instead of 24/96...while still "ok" the extra uplift/capability difference between 96 and 192 (with proper source) is a huge difference.also they should have supplied more info on how much "power" this DAC has I imagine because it is powered by USB alone it will not have a bunch of extra power to drive higher impedance headphones, also, just to point it out, while it seems "neat" and might offer a few things others do not, that is a hell of an asking price IMO when one is able to get stand alone DC/Amplifier for a fraction of the price.
because it is steel series likely they require their drivers (which are at best "passable" if they are not wanting to update themselves all the damn time lol...yes I had experience with steel series stuff)
Boxie - Tuesday, August 28, 2018 - link
Apart from chasing specs - what is the benefit you see in 96khz? The ability to play some ultrasonic frequencies that could potentially degrade your listening experience (if your headphones don't like it etc etc)?48khz is way more than enough frequencies (thanks to Nyquist - 24khz) for you to listen to, and (at least for me) about ~6khz of that I can't really hear anyway.
USB3 can supply ~ 0.9amp - which should be more than enough to make those higher frequencies irrelevant to you quickly enough :P
liteon163 - Tuesday, August 28, 2018 - link
24/96 doesn't need a driver on Windows, as the stock UAC2 driver covers it.I too prefer hardware that supports at least 24/192. While the extreme higher frequencies may largely be inaudible, the lower noise floor and higher dynamic range will benefit anyone with decent headphones. And the more capable hardware required to support 24/192 tends to make require higher quality components, which generally benefits ALL sound, except poorly encoded or highly compressed sources.
Veto - Tuesday, August 28, 2018 - link
Are you a bat???Frequencies above ~20kHz are not "largely inaudible" - they are completely inaudible. Furthermore reproducing ultrasound frequencies may introduce unwanted artifacts in the audible range due to non-linearities of the speakers etc..
Using sample frequencies of 96/192 ksps is marketing bullshit. And frankly, since when has "gaming" audio been synonymous with high-end Hifi anyway?
liteon163 - Tuesday, August 28, 2018 - link
Have you ever listened to a 24/192 FLAC of uncompressed music (and by this I mean music that a producer has not artificially compressed) with good headphones on quality hardware? It's not all about high frequency, it's also about being able to hear low level sound that would be obscured by chips unable to produce the dynamic range need to hear EVERYTHING.Moizy - Tuesday, August 28, 2018 - link
Gentlemen - 96kHz refers to the sampling rate, not the sound frequency. Obviously there is zero value in capturing 96kHz frequencies that your dog can't hear. That's not what 96kHz is in reference to.A better way to understand it is to think about 3D printing. Have you seen 3D prints done on cheap printers? You know how you can see the steps in the side of the object as it builds level by level on the 3D print, making the printed object looks crappy and a lot rougher in texture than the original design. It can't print a continuous object, instead it has to break it down into slices and then express it that way. Well, one difference between cheap 3D printers and nice ones is the nice ones can use really fine slices, so fine that if you stand at a reasonable distance you might not notice they're there and it approximates the original design.
This is what 48/96/192kHz is talking about. Your favorite song that includes audible sounds from 20Hz to 20kHz, how finely do you want the compressed, digitized "print" or copy of that song to slice it? The finer it's sliced, the better it sounds, to a point obviously. See "Sampling" on Wikipedia.
ZeDestructor - Wednesday, August 29, 2018 - link
That's not how Shannon-Nyquist works. Here, have some links and educate yourself before repeating stuff that's just plain wrong:https://people.xiph.org/~xiphmont/demo/neil-young....
https://www.youtube.com/watch?v=cIQ9IXSUzuM
tl;dr and tl;dw: sampling rate defines the maximum frequency you can record (specifically sampling rate/2 == max, so 44.1kHz sample rate of a CD gives you 22.05kHz max, for example). It has absolutely no bearing on "resolution" of sound, because we can already perfectly represent any sound any human can hear.
Moizy - Thursday, August 30, 2018 - link
Very interesting links, and they do cast doubt on my 3D printing analogy, but the links also teach sampling rate does not equal the sound frequency, which is what my overall point was. A 96kHz sampling rate does not mean the sound frequency will be extended out to 96kHz. Your first link had a link to an interesting article: http://lavryengineering.com/pdfs/lavry-sampling-th... . The Nyquist theory, from what I can read, describes the number of slices you'll need to perfectly describe a sound or frequency. Adding more slices adds nothing to the presentation. In this way, the analogy to screen resolution isn't terrible, in that we can reach a point with resolution that adding more pixels to a phone screen or a PC display adds zero value or additional data but comes at a great cost, in file size, in processing requirements, in hardware requirements, in power requirements. I'm not proposing a certain sampling rate, but I do know the sampling rate does not equal the sound frequency contained by a file. If that were the case, we could be happy with 24kHz sampling, but give that a try sometime and see how it works out for you compared to 48kHz.Diji1 - Wednesday, August 29, 2018 - link
That's nice but it's confirmation bias at work. You are not a magical person who doesn't suffer from it.Diji1 - Wednesday, August 29, 2018 - link
Just resampling alone introduces artifacts.Gasaraki88 - Tuesday, August 28, 2018 - link
No special drivers needed.jabber - Tuesday, August 28, 2018 - link
How many audio devices have a frequency output higher than 22Khz? Very few.I've seen hi-res 'buffs' going on about how hi-res is amazing then i ask what speakers they run it all through...oops. Bit rate probably helps more with resolution than frequency. They should have just upped Red Book to 24bit and left it at that. 18 or 20bit would have probably been fine.
ZeDestructor - Wednesday, August 29, 2018 - link
Human hearing is generally accepted to be 20Hz to 20kHz, though some argue that true golden ears can hit 21, or even 22kHz (I haven't heard of any yet, though...). It doesn't matter that there are plenty of speakers, DACs and amps that can render higher (albeit very non-linearly, and we know that cause we can measure it), if you can't hear it.Remember, you have to account for the noise floor - 20dB(SPL) in a studio, 30-50 dB(SPL) for more typical surroundings; and the threshold of pain - ~120dB(SPL). That means that the maximum theoretical dynamic range you can reach is 100dB. 16bits lets you encode 96dB of dynamic range without using any fancy tricks.
In practice, the human ear has much lower dynamic range because human hearing dynamically adjusts to compensate for loudness. Combine with not wanting music to be "too loud", one can reasonably conclude that the 96dB dynamic range of 16bit audio is more than enough to capture any human-perceptible audio. Until the end of time.
That said, there are reasons for having higher bit depths, and that would be mixing, production and mastering, where higher bit-depths just make things more convenient. You can, of course still work entirely in INT16, but you need to take some care with effects, clipping and distortion. But working in INT24, or FP32 (in particular) is just so much more convenient, so that's what you see in production.
Diji1 - Wednesday, August 29, 2018 - link
>They should have just upped Red Book to 24bit and left it at that. 18 or 20bit would have probably been fine.No that would have been stupid because the human ear is incapable of discerning higher than 16bit fidelity. It is literally mechanically impossible for the ear to discern greater fidelity, there is no higher quality.
Of course many people including audiophiles are going to explain that this is wrong because they don't understand sampling and because they don't understand that confirmation bias is automatic and unconscious.
Lord of the Bored - Wednesday, August 29, 2018 - link
"They should have just upped Red Book to 24bit and left it at that. 18 or 20bit would have probably been fine. "You mean HDCD? Yeah, that was a nice idea. Pity it never took off.
wrkingclass_hero - Tuesday, August 28, 2018 - link
Bit depth has a much larger impact than frequency. The audible difference between 96 and 192 Kilohertz is very minor. You'd be much better off increasing the bit depth to 32 bit floating point.Diji1 - Wednesday, August 29, 2018 - link
>You'd be much better off increasing the bit depth to 32 bit floating point.That would be pointless other than people thinking 32bit is higher quality when it isn't.
ZeDestructor - Tuesday, August 28, 2018 - link
Here, have some actual hard science on why you really don't need anything beyond 16bit, 44.1khz: https://people.xiph.org/~xiphmont/demo/neil-young.... .The ONLY place you need anything higher than that is when you're actually mastering or mixing audio. That would be handled internally by WASAPI/ALSA/OSS/PulseAudio/JACK, not by the DAC. For pure audio storage and output, 16/44.1 is all you need, although DVD and BR have pushed the standard a little up to 24bit/48kHz (somewhat unnecessary, but not actually harmful).
Gasaraki88 - Tuesday, August 28, 2018 - link
What a bunch of F***ers. I bought the bundle when it first came out and you can't use any other headphones with the DAC. Now they release one 3 months later that you can use any headphones. People don't buy their headphones, don't listen to the reviews.Brett Howse - Tuesday, August 28, 2018 - link
I imagine the standalone version is just going to come with a cable to plug in the 3.5mm headsets and SteelSeries sells their cables on their site. Basically I would expect to see the opposite of this appear on the site soon:https://steelseries.com/gaming-accessories/arctis-...
If that's what you're after and you don't see it in the coming months send me an email and I can reach out to their PR people.
PeachNCream - Tuesday, August 28, 2018 - link
It's nice the DAC is standalone now. I'm pretty excited about that because I can pair it up with any old set of earbuds or headphones without blowing $70+ on overpriced Arctis ones when I'm fine with a $10 throw away set off Amazon or out of the local Wally World.Sttm - Tuesday, August 28, 2018 - link
The DAC reminds me of the Soundblaster G5 I have, which I think is dying on me as sometimes it randomly disconnects and connects. I wonder if you can listen to both optical and PC audio at the same time on this one. That is the killer feature that has kept me from trying other dac/headphone amp solutions.denisfrancis - Wednesday, October 6, 2021 - link
Wow, these headphones are really cool designed and quality made.https://blogs.beingawaisali.com/blog/best-headphon...